To my knowledge we are not calling the g_free twice on the same place. 




In some places we do set the pointer to NULL after freeing it.  But not in all the places.





In most of the places we have the following code 


if (a_ptr != NULL)


    g_free(a_ptr);



Since the g_free does not set a_ptr to NULL the first check itself becomes invalid. If the same pointer is allocated using g_malloc (instead of g_malloc0) we get the same contents that was there before g_free!. The following example demonstrates that. So it is safer to set the pointer to NULL once the memory is free'd.






Creating a wrapper takes care of it all in one place. Check for NULL, free it, set it to NULL after freeing it. 



/* Use cc -o g_free_null g_free_null.c -lglib-2.0 */
#include <stdio.h>
#include <stdlib.h>

main()
{
    char *mem_ptrs = 0;
    int len = 124;
    printf("mem_ptrs pointer value at start is %p\n",mem_ptrs);
    printf("The g_malloc is NOT supposed to initialize the allocated memory\n");
    mem_ptrs = (char *)g_malloc(sizeof(char) * len);
    printf("The allocated string with g_malloc is %s and ptr is %p\n",mem_ptrs, mem_ptrs);
    strcpy(mem_ptrs,"Mohan");
    g_free(mem_ptrs);
    printf("mem_ptrs pointer value after g_free is %p\n",mem_ptrs);
    mem_ptrs = 0;
    printf("mem_ptrs pointer value after mem_ptrs = 0 is %p\n",mem_ptrs);

    printf("The g_malloc is NOT supposed to initialize the allocated memory\n");
    mem_ptrs = (char *)g_malloc(sizeof(char) * len);
    printf("The allocated string with g_malloc is %s and ptr is %p\n",mem_ptrs, mem_ptrs);
    g_free(mem_ptrs);
    printf("mem_ptrs pointer value after g_free is %p\n",mem_ptrs);
    mem_ptrs = 0;
    printf("mem_ptrs pointer value after mem_ptrs = 0 is %p\n",mem_ptrs);

    printf("The g_malloc0 is supposed to initialize the allocated memory\n");
    mem_ptrs = (char *)g_malloc0(sizeof(char) * len);
    printf("The allocated string with g_malloc0 is %s and ptr is %p\n",mem_ptrs, mem_ptrs);
    g_free(mem_ptrs);
    printf("mem_ptrs pointer value after g_free is %p\n",mem_ptrs);
    mem_ptrs = 0;
    printf("mem_ptrs pointer value after mem_ptrs = 0 is %p\n",mem_ptrs);


}




Output is

[root@dl380g7-1 bin]# ./g_free_null
mem_ptrs pointer value at start is (nil)
The g_malloc is NOT supposed to initialize the allocated memory
The allocated string with g_malloc is  and ptr is 0xc5b010
mem_ptrs pointer value after g_free is 0xc5b010
mem_ptrs pointer value after mem_ptrs = 0 is (nil)
The g_malloc is NOT supposed to initialize the allocated memory
The allocated string with g_malloc is Mohan and ptr is 0xc5b010
mem_ptrs pointer value after g_free is 0xc5b010
mem_ptrs pointer value after mem_ptrs = 0 is (nil)
The g_malloc0 is supposed to initialize the allocated memory
The allocated string with g_malloc0 is  and ptr is 0xc5b010
mem_ptrs pointer value after g_free is 0xc5b010
mem_ptrs pointer value after mem_ptrs = 0 is (nil)
















 On Tuesday, November 11, 2014 1:13 PM, Bryan Sutula <sutula@users.sf.net> wrote:
 




I don't understand why we are permitting code to call g_free twice on the same piece of memory.  It seems to me that this is a coding bug and should be addressed in the calling code.

The core dump is merely a symptom of a coding problem.  If you mask the core dumps, you will no longer get notified of the root problem.



---

** [bugs:#1862] Create wrap_g_free to set the freed memory to NULL**

**Status:** open
**3.6.0:** 3.6.0
**Labels:** HP c-Class Plugin 
**Created:** Tue Nov 11, 2014 06:24 PM UTC by dr_mohan
**Last Updated:** Tue Nov 11, 2014 06:24 PM UTC
**Owner:** Hemantha Beecherla

g_free call does not set the pointer to NULL after freeing the memory. This is troublesome as some other thread could potentially try to free the pointer again  leading to core dumps. Create a wrapper funtion to g_free which will free the memory and set the pointer to NULL.

Also g_malloc0 initializes the contents to 0 ater allocating it. There are places where g_malloc is used instead of g_malloc0. Modify those places also.


---

Sent from sourceforge.net because you indicated interest in <https://sourceforge.net/p/openhpi/bugs/1862/>

To unsubscribe from further messages, please visit <https://sourceforge.net/auth/subscriptions/>